christianschuler8989/asr_system_experiment_viewer_darija

A static web application for native-speaker review of ASR-system transcriptions, hosted on GitHub Pages # ASR Experiment Review Tool A static web application for native-speaker review of ASR-system transcriptions, hosted on GitHub Pages. Click here to check out the current prototype ## Project Structure ``` / ├── index.html # Main application entry point ├── app.js # Application logic ├── style.css # Styles ├── data.json # Aggregated transcription data (generated) ├── audio/ │ ├── audio_files_transcriptions/ # .gitignored input data │ └── audio_files/ # Audio files, one sub-folder per location │ ├── Casablanca/ │ ├── Tangier/ │ └── Oujda/ ├── scripts/ │ └── aggregate_asr_data.py # Generates data.json from raw data ├── README.md └── ROADMAP.md ``` ## UI Layout concept ``` ┌─────────────────────────────────────────────┐ │ [Casablanca ▾] [File 01 ▾] ▶ Play │ ├─────────────────────────────────────────────┤ │ Spectrogram + playhead │ ← wavesurfer.js ├─────────────────────────────────────────────┤ │ moulsot_v0_3 │ وقال الرئيس... │ ← selectable text │ qwen3_asr │ قال الرئيس... │ │ whisper_darija │ و قال... │ │ ... │ ├─────────────────────────────────────────────┤ │ [Annotation panel — per file notes] │ │ [Export Annotations JSON] │ └─────────────────────────────────────────────┘ ``` ## Setup **Prerequisites:** Python 3.8+ (for the aggregation script and local server). No other installation required — the app runs entirely in the browser with no build step. --- **1. Place input files** Audio files, transcriptions, and TextGrids belong under `audio/` as follows: ``` audio/ audio_files/ / .wav audio_files_transcriptions/ / __ .TextGrid (optional, per-system) transcriptions- .csv ``` Supported locations: `Casablanca`, `Oujda`, `Tangier`. --- **2. Regenerate `data.json` (only needed …